This paper presents preliminary findings and is being distributed to economists

and other interested readers solely to stimulate discussion and elicit comments.
The views expressed in this paper are those of the authors and are not necessarily
reflective of views at the Federal Reserve Bank of New York or the Federal
Reserve System. Any errors or omissions are the responsibility of the authors.
Federal Reserve Bank of New York
Staff Reports


Federal Reserve Tools for Managing Rates and
Reserves


Antoine Martin
James McAndrews
Ali Palida
David Skeie



Staff Report No. 642
September 2013
Federal Reserve Tools for Managing Rates and Reserves
Antoine Martin, James McAndrews, Ali Palida, and David Skeie
Federal Reserve Bank of New York Staff Reports, no. 642
September 2013
JEL classification: E42, E43, G12, G20









Abstract

Monetary policy measures taken by the Federal Reserve as a response to the 2007-09 financial
crisis and subsequent economic conditions led to a large increase in the level of outstanding
reserves. The Federal Open Market Committee (FOMC) has a range of tools to control short-term
market rates in this situation. We study several of these tools, namely, interest on excess reserves
(IOER), reverse repurchase agreements (RRPs), and the term deposit facility (TDF). We find that
overnight RRPs (ON RRPs) may provide a better floor on rates than term RRPs because they are
available to absorb daily liquidity shocks. Whether the TDF or RRPs best support equilibrium
rates depends on the intensity of interbank monitoring costs versus balance sheet costs,
respectively, that banks face. In our model, using the RRP and TDF concurrently may most
effectively stabilize short-term rates close to the IOER rate when such costs are rapidly
increasing.

Key words: monetary policy, fixed-rate full allocation overnight reverse repurchases, term
deposit facility, interest on excess reserves, FOMC, banking















_________________
Martin, McAndrews, Palida, Skeie: Federal Reserve Bank of New York (e-mail:
antoine.martin@ny.frb.org, jamie.mcandrews@ny.frb.org, ali.palida@ny.frb.org,
david.skeie@ny.frb.org). The authors thank Alex Bloedel for research assistance and seminar
participants at the Federal Reserve Bank of New York and the Board of Governors for helpful
comments. The views expressed in this paper are those of the authors and do not necessarily
reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System.
1 Introduction
This paper studies new monetary policy tools for managing short-term market rates.
The tools we consider are interest on excess reserves (IOER), reverse repurchase
agreements (RRPs) with a wide range of market participants, and the term deposit
facility (TDF).
The Federal Reserve responded to the 2007-09 nancial crisis and its aftermath
with a wide range of monetary policy measures that dramatically increased the
supply of reserves. In part, this has led the federal funds rate, and other money
market interest rates, to be more variable than before the crisis. In October 2008,
the Federal Reserve began paying IOER to depository institutions (DIs). Since then,
money market rates have consistently remained below the IOER rate. In June of
2011, the Federal Open Market Committee (FOMC) announced a strategy for the
possible use of new tools geared towards inuencing short-term market interest rates
and keeping them close to the IOER.
2
In August 2013, the FOMC also announced a
xed-rate, full-allotment overnight RRP (ON RRP) as one of these potential tools.
3
IOER is paid to DIs holding reserve balances at the Federal Reserve.
4
Term
and ON RRPs provide a wide range of bank and non-bank counterparties with the
opportunity to make the economic equivalent of collateralized loans to the Federal
Reserve. The TDF is a facility oered to DIs, who are eligible to earn interest on
balances held in accounts at the Federal Reserve, that allows them to hold deposits
for longer term for an interest rate generally exceeding the IOER. An institutional
background and explanation of these tools is provided in Section 2.
We develop a general equilibrium model of banking and money markets to study
how the Federal Reserve can manage short-term market rates and the large level of
reserves on its balance sheet using these tools. Our model extends Martin, McAn-
drews, and Skeie (2013) (hereforth referred to as MMS) to include two separate
banking sectors, randomized relocation liquidity shocks occurring in an interim pe-
riod, and interbank lending frictions. The model provides a framework within which
to study the eectiveness of IOER, the term and ON RRP and the TDF in support-
ing interest rates, and provides insight into the economic mechanisms that determine
the equilibrium rates and quantities.
In our model, when banks are subject to balance sheet costs as in MMS, a with-
drawal by depositors (caused by a "liquidity shock") must be redeposited in other
2
Source: http://www.federalreserve.gov/monetarypolicy/les/fomcminutes20110622.pdf
3
Source: http://www.federalreserve.gov/newsevents/press/monetary/20130821a.htm
4
IOER diers from interest on reserves (IOR) in that IOER is paid to reserve holdings in excess
of the reserve requirement.
1
banks.
5
Those banks will react to the exogenous additional deposits and unplanned
balance sheet expansion by lowering their deposit rates. Alternatively, households
can hold government bonds, which would shield households from having to resort
to short-term deposits with depressed rates, and thus deposits bear a liquidity-risk
premium. In equilibrium, the liquidity shock leads to downward pressure on both
deposit rates and government bond yields.
Liquidity shocks aect banks asset returns as well. When a bank faces stochastic
withdrawals by its depositors, liquid reserves serve as a buer, allowing the bank
to fund these withdrawals with accumulated reserves. However, when outstanding
reserves are held in insucient quantities, in equilibrium, a banks funds are tied up
in illiquid assets and limit the banks ability to accommodate the withdrawal shock
on its own. In this situation, the bank must resort to borrowing on an interbank
market. When interbank lending frictions, such as monitoring costs, are present, the
interbank loan rate increases, reecting the costs of the frictions in interbank lending.
When there is a positive probability of experiencing suciently large withdrawals,
banks have a stronger incentive to hold their assets as liquid reserves, rather than
tie them up in illiquid assets such as loans to rms. For illiquid assets to be held in
positive amounts, they must earn a premium over reserve holdings in equilibrium.
The Federal Reserve has the ability to aect these spreads through its choice
of the quantity of reserves in the banking system and the size and implementation
of other central bank facilities that provide a broad array of tools for the Federal
Reserve.
IOER is oered to banks. It inuences deposit rates as it represents the riskless
return on an invested deposit at the Federal Reserve. It also sets a short-term
reservation rate in the interbank market at which banks should not lend below.
Federal Reserve RRPs provide an additional investment source for an expanded
set of intermediaries, such as money market mutual funds (MMFs). RRPs can be
used to raise rates by diverting deposits away from banks and into Federal Reserve
RRP non-bank counterparties. This supports deposit rates by reducing banks bal-
ance sheet size. Balance sheet size may be costly because of capital requirements,
leverage ratios, FDIC deposit insurance assessments, and other balance sheet costs.
RRPs may be either xed-rate, full-allotment or xed-quantity, and may be term
or overnight. Because the overnight rate reects daily liquidity shocks, the xed-rate,
full-allotment ON RRP is the most eective facility for setting a xed reservation
rate for those intermediaries; term or xed-quantity RRPs cannot achieve the same
level and stability of interest rates.
5
For simplicity, we will refer to DIs as banks in our framework.
2
In comparison to the RRP, the TDF absorbs liquid reserves without reducing
the size of bank liabilities and increases bank asset returns more directly. If RRPs
and the TDF are used in suciently large size, they can re-establish the interbank
market by reducing the size of liquid reserves used to ward o liquidity shocks and
can raise equilibrium bank asset returns. We nd that utilizing both the TDF and
the RRP together may support rates most eectively if both bank balance sheet
costs and interbank lending frictions are large enough.
Our paper ts broadly into the existing literature on monetary policy implemen-
tation, IOER, and reserves. Poole (1970) shows that the eectiveness of an interest
rate-change policy versus a money stock-change policy is not well determined and
depends on parameter values, but a combination policy is always weakly superior
to either of the two used alone. Ennis and Keister (2008) provide a general frame-
work for understanding monetary policy implementation with IOER. They show
that IOER can help implement a oor on market rates and allows the Federal Re-
serve to keep interest rates closer to target rates. MMS focuses on the eects of
excess reserves on ination, interest rates, and investment. They nd that these
parameters are largely independent of bank reserve holdings unless external fric-
tions are present. Bech and Klee (2011) analyze the federal funds market in the
presence excess reserves. They argue that since government-sponsored enterprises
(GSEs) do not have access to IOER, they have lower bargaining power and trade at
rates lower than IOER, thus resulting in the observed IOER-federal funds eective
rate spread. Kashyap and Stein (2012) show that, with both IOER and reserve
quantity control, the central bank can simultaneously mantain price stability and
address externalities resulting from short-term debt issuance. The current paper is
the rst to analyze the additional Federal Reserve tools and their eectiveness in
controling short-term money market rates and managing Federal Reserve liabilities.
The paper is organized as follows: Section 2 explains institutional details on the
Federal Reserves monetary policy before, during, and after the 2007-09 nancial
crisis and provides descriptions of IOER, RRPs and the TDF. Section 3 presents
and solves the benchmark model. Section 4 incorporates the RRP and the TDF
into the benchmark model and analyzes their equilibrium results and eectiveness.
Section 5 concludes. Proofs of some propositions and all gures are in the Appendix.
3
2 Institutional Background
Prior to the nancial crisis of 2007-2009, the Federal Reserve closely controlled
the supply of reserves in the banking system through its open market operations
(OMOs). In an OMO, the Federal Reserve buys or sells assets, either on a tempo-
rary basis (using repurchase agreements) or on a permanent basis (using outright
transactions), to alter the amount of reserves held in the banking system.
6
For
example, purchasing Treasuries will increase the amount of reserves in the system.
By adjusting the supply of reserves in the system, the open market trading desk
(the Desk) at the Federal Reserve Bank of New York (NY Fed) could inuence the
level of the federal funds rate, the rate at which DIs lend reserves to each other. DIs
in the US are required to maintain a certain level of reserves, proportional to specied
deposit holdings, which, in addition to precautionary demand for reserve balances,
creates a demand curve for reserves. The interest rate at which the demand and
the supply curves intersect increases when the Desk reduces the supply of reserves,
for example.
7
Through arbitrage, the level of the federal funds rate inuences other
short-term money markets rates.
In response to the 2007-09 nancial crisis and subsequent economic downturn,
monetary policy measures included large-scale lending to provide liquidity to -
nancial institutions, and large-scale asset purchases through the large-scale asset
purchase program (LSAP) to stimulate the economy by lowering longer term inter-
est rates.
8
This facilitated a very large increase in the supply of reserves.
9
Moreover,
in December 2008, the FOMC lowered the target federal funds rate to a range of 0
to 25 basis points, its eective zero bound, to help stimulate the economy.
10
The eective federal funds rate, a weighted average of federal funds trades
arranged by brokers, remained below 25 basis points, as shown in gure 1.
11
Figure 1
highlights that the federal funds rate uctuated closely with other short-term money
market rates, including the overnight Eurodollar rate and the overnight Treasury
repo rate. These rates are seen to be typically decreasing in the level of reserves.
6
Assets eligible for OMOs are Treasuries, agency debt, and agency mortgage-backed securities
(MBS).
7
See Ennis and Keister (2008) and Keister, and Martin, and McAndrews (2008) for a more
detailed introduction to traditional Federal Reserve monetary policy and OMOs
8
The LSAPs are sometimes referred to as "quantitative easing" (QE)
9
See Gagon, Raskin, Remanche, and Sack (2010) for more information on the LSAPs.
10
Source: http://www.federalreserve.gov/monetarypolicy/les/fomcminutes20081216.pdf
11
One common explanation for this is the current large presence of GSE lending in the federal
funds market. GSEs are not eligible for IOER and tend to lend at rates below 25 basis points (see
Beck and Klee (2011)).
4
See gure 1
In light of the LSAPs and the large expansion of the balance sheet, the Federal
Reserve has been preparing a variety of tools to ensure that short-term rates can
be raised when needed. IOER has been used as one of these tools since October
2008; however two of these tools, RRPs with an extended range of counterparties,
and the TDF, have not been implemented in large-value facilities as of yet. In
a 2009 speech, NY Fed President William Dudley, referred to the RRP and the
TDF as the suspenders that will support IOER, i.e. the belt, in allowing the
Federal Reserve retain control of monetary policy.
12
In August 2013, the FOMC
announced potential use of an additional tool, the ON xed-rate, full-allotment
RRP.
2.1 Interest on Excess Reserves
To manage short-term rates in the face of large excess reserves, the Federal Reserve
began to pay DIs IOER in October 2008. IOER diers from interest on reserves
(IOR) in that IOER is paid to reserve holdings in excess of the reserve requirement.
The Financial Services Regulatory Relief Act of 2006 originally granted the Federal
Reserve the ability oer IOER. However, the original authorization was only ap-
plicable to balances held by DIs starting October 2011. The Emergency Economic
Stabilization Act of 2008 accelerated the start date to October 2008.
13
The interest owed to a balance holder is computed over a maintainence period,
typically lasting one to two weeks depending on the size of the DI. Interest pay-
ments are typically credited to the holders account about 15 days after the close
of a maintainence period.
14
IOER was rst oered in October of 2008 at 75 basis
points, but is currently at 25 basis points where it has been since December 2008.
Institutions that are not DIs are not eligible to earn IOER.
15
2.2 Reverse Repurchase Agreements
An RRP is economically equivalent to a collateralized loan made to the Federal Re-
serve by a nancial institution. RRPs have historically been used, though somewhat
12
Source:http://www.newyorkfed.org/newsevents/speeches/2009/dud090729.html
13
Source:http://www.federalreserve.gov/newsevents/press/monetary/20081006a.htm
14
Source:http://www.federalreserve.gov/newsevents/press/monetary/monetary20081006a2.pdf
15
Source:http://www.federalreserve.gov/newsevents/press/monetary/20081006a.htm
5
infrequently, by the Federal Reserve in the conduct of monetary policy, arranged
with a set of counterparties called primary dealers.
16
In October 2009, the Federal Reserve announced that it was considering oering
RRPs on a larger scale to an expanded set of counterparties.
17
The expanded set of
counterparties include DIs as well as non-DIs, such as MMFs, GSEs, and dealers,
increasing both the number and the type of Federal Reserve counterparties.
18
In
addition, in August 2013 the Federal Reserve announced it would further study the
potential for adopting a xed-rate, full-allotment ON RRP facility.
19
In his Septem-
ber 2013 speech, President Dudly discussed this new facility as a way to support
money market rates by allowing counterparties a exible amount of investment at
a xed rate when needed.
20
RRPs do not change the size of the Federal Reserves balance sheet, but modify
the composition of its liabilities. Indeed, each dollar of RRPs held by counterparties
reduces one-for-one reserves held by DIs.
21
The Federal Reserve Bank of New York has held numerous small-scale temporary
operational exercises of RRPs for eligible counterparties starting in the fall of 2009.
22
Most recently, small-scale operational exercises have been held in April, June and
August of 2013. These operations were limited in terms of their overall size (less
than $5 billion) and were focused on ensuring operational readiness on the part of
the Federal Reserve, the triparty clearing banks, and the counterparties.
23
While
the Desk has the authority to conduct RRPs at maturites ranging from 1 business
day (overnight) to 65 business days, the operational exercises thus far have typically
ranged fromovernight to 5 business days, with several of the August 2013 operational
exercises consisting of overnight RRPs. Overnight RRPs were typically settled the
day after, however overnight RRPs with same day settlement were oered in August
2013.
At the September FOMC meeting, the committee authorized the Desk to imple-
16
See http://www.newyorkfed.org/markets/primarydealers.html.
17
Source:http://www.newyorkfed.org/newsevents/news/markets/2009/an091019.html
18
A full list of current eligible counterparties is available at
http://www.newyorkfed.org/markets/expanded_counterparties.html
19
Source:http://www.federalreserve.gov/monetarypolicy/les/fomcminutes20130731.pdf
20
Source:http://www.newyorkfed.org/newsevents/speeches/2013/dud130923.html
21
See the New York Fed page on RRPs for more information:
http://data.newyorkfed.org/aboutthefed/fedpoint/fed04.html
22
See the New York Fed page on temporary operations for a listing of recent RRP excercises:
http://www.newyorkfed.org/markets/omo/dmm/temp.cfm
23
These excercises were approved by the FOMC in November 2009.
See:http://www.federalreserve.gov/monetarypolicy/les/fomcminutes20091216.pdf
6
ment xed-rate RRP exercises with per-counterparty bid caps to limit the aggregate
size of the facility. In comparison to previous exercises, these excercises should bet-
ter simulate the xed-rate, full-alottment facility, which was discussed in the July
2013 FOMC minutes.
24
2.3 Term Deposit Facility
The TDF is another policy tool that can reduce reserves but it is available only
to DIs.
25
The TDF was approved in April 2010, following the approval of amend-
ments to Regulation D (Reserve Requirements of Depository Institutions), allowing
Federal Reserve Banks to oer term deposits to institutions eligible to earn interest
on reserves.
26
Small value temporary operational exercises of term deposits have
occurred since June 2010, and recent small value operational exercises have been
held in March, May, and July, and September of 2013.
27
As was the case for RRPs, the TDF does not change the size of the Federal
Reserves balance sheet, but alters the composition of its liabilities. Reserves used
to nance purchases of term deposits are unavailable to DIs until the term deposit
matures. The TDF therefore directly absorbs reserves when banks substitute reserve
holdings for TDF holdings.
3 Benchmark Model
3.1 Agents
The economy lasts three periods t = 0. 1. 2 and consists of two sectors, i = 1. 2.
which are partially segmented. Each sector contains three agents: a bank, a rm,
and a risk-neutral household. In addition, a nancial intermediary that we associate
with an MMF operates across both sectors. The central bank and the government
issue liabilities but do not behave strategically.
At date 0, households in each sector receive an endowment (1
i
) that can be held
in the form of deposits in the bank of their sector (1
i
), or in MMF shares (1). No
other agent has an endowment.
24
Source:http://www.newyorkfed.org/markets/rrp_faq.html
25
Source:http://www.federalreserve.gov/newsevents/press/monetary/20100430a.htm
26
Source:http://www.federalreserve.gov/newsevents/press/monetary/20100430a.htm
27
Source:http://www.frbservices.org/centralbank/term_deposit_facility_archive.html
7
The supply of reserves and government bonds are set exogenously and denoted
by M and 1, respectively. The interest paid on reserves is set exogenously and
denoted by 1
M
paid each period, while the interest paid on government bonds is
determined in equilibrium and denoted 1
B
.
Banks take deposits from households and can invest them either in loans to
the rm from the same sector (1
i
) or in reserves at the central bank (`
i
), with
`
1
+ `
2
= M. Note that only banks can hold reserves. Reserves are injected by
purchasing bonds, so the quantity of bonds held by the central bank, 1
CB
, is equal
to the supply of reserves M.
Firms borrow from banks and nance projects with a concave and strictly in-
creasing production function, with marginal return given by :
i
(1
i
). The rms
output is sold as consumption goods to households at date 2.
28
The MMF can sell shares to households and invest in government bonds. We
denote the MMFs holding 1
H
.
Banks, rms, and the MMF are prot maximizers, while households seek to max-
imize consumption. There are centralized markets for goods, bonds, and reserves,
which imply they have common prices and returns across sectors. However, the
returns on the other assets in the economy can vary across sectors. We abstract
from credit risk, as the focus of the paper is the use of monetary policy tools in a
stable, non-crisis environment.
3.2 Timeline
At t = 0 assets are traded in both sectors. The household of sector i deposits 1
0
i
in
the local bank and invests 1
0
i
in MMF shares. Banks accept deposits, hold reserves,
and lend to rms at a rate 1
L
. The MMF sells share and purchases bonds.
At t = 1. a liquidity shock hits one of the two sectors. The probability that
sector i is hit is
1
2
for i = 1. 2. In the sector aected by the shock, a fraction ` of
households must withdraw their deposits from their bank because they relocate to
the other sector. In the benchmark case, the only option for relocated households
is to deposit in the bank of the sector they moved to.
Banks can use reserves to meet withdrawals. Reserve have a face value of 1
M
at
t = 1 per unit held at t = 0. If a bank does not have enough reserves, it can borrow
1 in the interbank market at an interest rate of 1
I
. We assume interbank lending
frictions in the form of a strictly increasing and convex cost, ,
i
(1), for the lending
28
Note that uppercase variables denote nominal values while lowercase variables denote real
values. Also subscripts always represent the sector.
8
bank, which represents monitoring costs. We assume that IOER is constant so that
a unit of reserve held from t = 1 to t = 2 is also 1
M
.
At t = 2, rms sell their output to households at a price of 1 per unit, households
consume the goods they purchase, and rms repay their loans to banks.
We assume for simplicity that deposits made at t = 0 can be withdrawn for a
return of 1 at period t = 1. Deposits not withdrawn at t = 1 yield a return of 1
D0
,
while deposits that were withdrawn and re-deposited yield a return of 1
D1
.
To facilitate analysis we choose a somewhat stylized structure of the MMF. We
choose to model MMF shares as oering a competitive return of 1
F0
for those sold
in t = 0 and 1
F1
for those oered in t = 1. Thus, we consider shares of the MMF
oered in dierent periods as investments in dierent funds.
A key friction is the presence of balance sheet costs for banks. Balance sheet
costs were introduced in MMS, with each bank bearing an exogenous cost that is
increasing in the size of their balance sheet, with marginal real cost given by c
i
(1).
Balance sheet costs are motivated by the analysis of market observers. For ex-
ample, interbank broker Wrightson ICAP (2008) voiced concerns that large reserves
could clog up bank balance sheets. Furthermore, as MMS explains, banks tended
to reduce the size of their balance sheets during the recent crisis, in line with the
presence of balance sheet costs. Evidence for this cost is also suggested by gure
1. We observe that the quantity of reserves is clearly negatively correlated with all
of the bank deposit and related short-term money market rates plotted. The table
below lists these correlation coecients.
29
Rate Correlation
Federal Funds Eective -.59
O/N Eurodollar -.57
4 Week T-Bill -.53
In MMS, this negative correlation is explained by balance sheet frictions bearing
exogenous costs on banks, which in equilibrium are pushed onto depositors. Thus,
when reserves, and consequently bank balance sheets, are large, the resulting fric-
tions are imposed on depositors through a lower deposit rate. Possible explanations
for these balance sheets costs include capital requirements, leverage ratios, and FDIC
deposit insurance assessments applied to all non-equity liabilities.
30
In July 2013,
29
Source: Federal Reserve Board H.15 report and H.4.1 re-
port: http://www.federalreserve.gov/releases/h15/update/ and
http://www.federalreserve.gov/Releases/h41/
30
Federal Reserve Bank of New York President Dudley states that to the extent that
9
the Federal Reserve and the FDIC proposed a new rule to strengthen leverage ratios
for the largest, most systemically important banks. Under the proposed rule, bank
holding companies with more than $700 billion in consolidated total assets would
be required to maintain a tier 1 capital leverage of 5 percent, 2 percent above the
minimum supplementary leverage ratio of 3 percent. Such proposals suggest that
balance sheet costs may be relevant especially in the near future.
31
We assume that this cost is aggregated across periods; that is, if 1
0
deposits
were issued in t = 0 and 1
1
deposits were issued in t = 1. then the total cost for
the bank due to balance sheet costs is:
1
"
Z
D
0
0
c
i
(1)d1 +
Z
D
0
+D
1
D
0
c
i
(1)d1
#
(1)
At period t = 2. households pay a lump-sum tax (t) such that the government
maintains a net balanced budget.
3.3 Optimizations
In this section we describe each agents optimization. Firms seek to maximize prots
obtained from sales of real goods in t = 2. Thus, the rm in sector i solves:
max
L
i
1
Z
L
i
0
:
i
(
b
1)d
b
1 1
L
i
b
1
The MMF maximizes prots and solves:
max
B
H
;F
0
1
B
1
H
1
F0
1
0
:.t. 1
H
= 1
0
The MMF simply arbitrages, in the bond market, the capital obtained from selling
their shares. They earn the spread between the total bond return and the claims
they must pay out for all shares in t = 2.
Households and banks must take into account the liquidity shock. The household
of sector i solves:
max
D
0
i
;F
0
i
1
2
(
1
D0
i
(1 `)1
0
i
+1
D1
`1
0
i
)
1
) + (
1
2
)(
1
D0
i
1
0
i
1
) +
1
F0
1
0
i
t
1
:.t. 1
0
i
+1
0
i
= 1
the banks worry about their overall leverage ratios, it is possible that a large increase
in excess reserves could conceivably diminish the willingness of banks to lend. Source:
http://www.newyorkfed.org/newsevents/speeches/2009/dud090729.html
31
Source:http://www.federalreserve.gov/newsevents/press/bcreg/20130709a.htm
10
The rst term in the objective function represents the expected return on deposits
from a household in the sector aected by the shock. With probability ` the house-
hold must relocate, withdraw 1
0
i
from its bank, and can redeposit in the bank of
the other sector. In such a case, the household earns 1
D1
on its deposit in the
sector it moved to. If the household does not need to relocate, then it earns 1
D0
i
on
its deposit. The household also earns 1
F0
1
0
i
for its investment in the MMF, and
pays t in taxes, regardless of whether it must relocate. Since the household values
real consumption, all nominal terms are divided by the price level in t = 2. The
constraint simply states that total household investment in t = 0 must equal total
endowment.
The banks optimization is similarly given by:
max
L
i
;M
i
D
0
i
;D
1
i
;I
1
2
(1
M
max(1
M
`
i
`1
0
i
. 0) 1
I
max(0. `1
0
i
1
M
`
i
)
(1 `)1
D0
i
1
0
i
1
Z
D
0
i
0
c
i
(
b
1)d
b
1)
+
1
2
(1
M
(1
M
`
i
+1
1
i
1
i
) +1
I
1
i
1
Z
I
i
0
,
i
(
b
1)d
b
1 1
D0
i
1
0
i
1
D1
1
1
i
1
Z
D
0
i
+D
1
i
0
c
i
(
b
1)d
b
1) +1
L
i
1
i
:.t. 1
i
+`
i
= 1
0
i
1
M
`
i
+1
1
i
1
i
0
With probability
1
2
, bank i is hit with the liquidity shock and must pay out `1
0
i
in deposits. Any reserves in excess of `1
0
i
can be reinvested for a return of 1
M
and the bank earns 1
M
(1
M
`
i
`1
0
i
). If outstanding reserves do not exceed `1
0
i
.
the dierence must be satised using interbank loans and the bank must pay out
1
I
(`1
0
i
1
M
`
i
) to the other bank. The bank must also pay out (1 `)1
D0
i
1
0
i
at date 2, and bear the balance sheet costs 1
Z
D
0
i
0
c
i
(
b
1)d
b
1). If the bank is not
hit by the shock, it receives additional deposit at t = 1 and may be requested to
provide an interbank loan. The payo earned on reserves held from t = 1 to t = 2 is
1
M
(1
M
`
i
+1
1
i
1
i
). The payo on the interbank loan is 1
I
1
i
1
Z
I
i
0
,
i
(
b
1)d
b
1. The
bank must also pay out its deposit liabilities, (1
D0
i
1
0
i
+ 1
D1
1
1
i
), and its balance
sheet cost 1
Z
D
0
i
+D
1
i
0
c
i
(
b
1)d
b
1). In either case, the bank will earn the same return
on loans, 1
L
i
1
i
. The rst constraint says that, since banks have no initial capital,
t = 0 deposits must equate with the banks t = 0 asset holdings 1
i
+`
i
. The second
constraint is to ensure that the bank does not make more interbank loans than they
11
have outstanding liquidity, 1
M
`
i
+1
1
i
.
3.4 Equilibrium Analysis
For simplicity, we consider standard market security instruments; we use general
equilibrium as our solution concept. In particular, an equilibrium in this economy
is a set returns, 1
D0
i
. 1
D1
. 1
B
. 1
F0
. 1
F1
. 1
L
i
. and 1
I
. and a t = 2 price level 1. such
that all markets clear at the agents optimizing levels of investment and consump-
tion.
We focus on an ex-ante symmetric case where both sectors and all agents are
identical. Therefore we assume that for all households i in both sectors we have
1
i
= 1. Furthermore for both sectors 1 and 2. we have:
:
1
(1) = :
2
(1) = :(1)
c
1
(1) = c
2
(1) = c(1)
,
1
(1) = ,
2
(1) = ,(1).
We also assume the standard regularity conditions:
:(1) 0. :
0
(1) < 0. :(0) = 1. :(1) = 1
c(1) 0. c
0
(1) 0. c(0) = 0. c(1) = 1
,(1) 0. ,
0
(1) 0. ,(0) = 0. ,(1) = 1.
Furthermore, we examine equilibria where the quantity of reserves and bonds
are symmetric across the two sectors at t = 0. That is, `
i
=
M
2
and 1
0
i
=
BM
2
for
i = 1. 2.
32
This implies that:
1
0
1
= 1
0
2
= 1
1 M
2
(2)
1
1
= 1
2
= 1
1
2
(3)
Since all the quantities at t = 0 are identical in equilibrium, we can drop the
subscripts. By symmetry, and because of the common good market, 1
D0
1
= 1
D0
2
and
32
Note that M
i
represents the amount of reserves held by the bank of sector i at t = 0. At t = 1
bank is outstanding reserves may change in response to liquidity shocks and/or interbank lending.
12
1
L
1
= 1
L
2
for both sectors in equilibrium. Therefore we can also drop the subscripts
on all rates and talk of one equilibrium rate for loans and deposits. It is easy to
show that only the bank from the sector that did not experience the shock oers
interbank loans. To further simplify notation, we assume that the bank experiencing
the shock does not receive deposits at t = 1.
33
Therefore, the equilibrium rates 1
D1
and 1
I
refer only to the returns oered in t = 1 for the non-shocked bank and we
can drop the subscripts on 1
1
i
and 1
i
.
We can now turn to the determination of equilibrium rates and quantities. The
level of the central banks supply of reserves plays a large role in the rates determined
in equilibrium. We rst dene M, the supply of sreserve above which banks have
enough reserves to fully fund potential withdrawals at t = 1:
M
`
1
M
`
(21 1) (4)
Propositions 1 and 2 characterize the equilibrium of this model for M Mand
for M< M.
Proposition 1 For M M a unique competitive equilibrium is given by (1)-(8):
1. 1 = 1
B
2
2. 1
L
= (1
M
)
2
3. 1
0
= 1
BM
2
. 1
1
= `1
0
. 1
0
=
BM
2
. 1
1
= 0
4. 1
D0
=
2
2
[(1
M
)
2

R
M
2
1(
1
2
c(1
0
)+
1
2
c(1
0
+1
1
))]. 1
D1
= 1
M
1c(1
0
+1
1
).
1
F1
1
D1
5. 1
F;0
= 1
B
=

2
1
D1
+ (1

2
)1
D0
6. 1 = 0
7. 1
I
= 1
M
+1,(0) = 1
M
8. 1 =
(R
M
)
2
r(E
B
2
)
33
We could also prove that there would be no t = 1 deposits for the shocked bank if depositors
could not withdraw and deposit into the same bank and non-shocked depositors could not relocate
to the other sector at t = 1:
13
Proof. See the Appendix.
Item 1 and the rst equality under item 3 show that the amount of loans and
deposits are pinned down by the endowment, the supply of government bonds, and
the supply of reserves, which are exogenously xed. Item 2 must hold for banks to
invest in both loans and reserves. Item 3 states that t = 1 deposits are equal to the
amount withdrawn by relocating depositors and that total investment in the MMF
is equal to the amount of government bonds remaining after the central bank has
completed its purchases. Like in MMS, balance sheet costs drive a wedge between
the deposit rate and IOER, as shown in item 4. The wedge increases with the size
of the balance sheet costs and, thus, the size of deposits. Item 5 says that the return
oered by the MMF, which is equal to the return on bonds, must be equal to the
expected return on deposits for depositors to invest in both the bank and the MMF.
Since M M, the interbank market is inactive, as noted in item 7. Banks have
enough liquidity to accommodate any potential payment shock and, thus, have no
incentive to engage in costly interbank borrowing. Finally, item 7 states that the
interbank return is equal to the IOER when there is no interbank lending.
When there is no shock (` = 0), proposition 1 is the case of moderate balance
sheet cost in MMS. Figure 2 represents the equilibrium graphically.
See gure 2
Proposition 2 characterizes the equilibrium when liquidity is not sucient to
cover withdrawals at date 1 and banks must use the interbank market. To facilitate
the analysis, we impose a small regularity condition:
:(1
1
2
)
1
2
1
M
,(`(1
1 M
2
) 1
M
`) (5)
Equation (5) states that the marginal return on production must be suciently large
compared to the marginal real interbank cost at the equilibrium loan and deposit
level.
Proposition 2 If (5) holds and M< M, a unique competitive equilibrium is given
by (1)-(8). If (5) does not hold, then no equilibrium with a nite, positive price level
and full redeposits at t = 1 exists.
1. 1 = 1
B
2
2. 1
L
= (1
M
)
2
+
1
2
(1
M
)
1=2
1,(1) (1
M
)
2
3. 1
0
= 1
BM
2
. 1
1
= `1
0
. 1
0
=
BM
2
. 1
1
= 0
14
4. 1
D0
=
2
2
[(1
M
)
2

R
M
2
+ 1,(1)(
R
M
2


2
) 1(
1
2
c(1
0
) +
1
2
c(1
0
+ 1
1
))].
1
D1
= 1
M
1c(1
0
+1
1
). 1
F1
1
D1
5. 1
F0
= 1
B
=

2
1
D1
+ (1

2
)1
D0
. 1
H
=
BM
2
6. 1 = `1
0
1
M
` 0
7. 1
I
= 1
M
+1,(1) 1
M
8. 1 =
(R
M
)
2
r(L)
1
2
R
M
f(I)

(R
M
)
2
r(L)
Proof. See the Appendix.
Items 1, 3, and 5 are the same as in proposition 1. Item 6 shows that banks
use the interbank market because liquidity is scarce. As is shown in gure 3 below,
and implicitly in item 7 of the proposition, the interbank rate is decreasing in the
amount of reserves and is always above the IOER.
See gure 3
Investing in loans rather than reserves means potentially having to borrow in the
interbank market, which is costly since 1
I
1
M
. Hence, the return on loans must
exceed the IOER rate, as shown in gure 4 and in item 2 of the proposition. Also,
since 1
L
is no longer pinned down at the IOER, item 8 shows that the equilibrium
price level is increasing in the volume of interbank loans and, thus, decreasing with
the level of reserves.
See gure 4
Given the presence of balance sheet costs, it is reasonable to believe that 1
D1
<
1
D0
. This is because t = 1 deposits are added to more congested balance sheets
than t = 0 deposits. The t = 1 deposits increase the balance sheet cost of the bank
receiving these deposits and the depositors have a perfectly inelastic demand for
these deposits. In proposition 1 we will have 1
D1
< 1
D0
when:
(1
M
)
2
1
M

1
2
(c(1
0
) (1 `)c(1
0
+1
1
)) (6)
and similarly for proposition 2:
(1
M
)
2
+1,(1)(
1
M
2

`
2
) 1
M

1
2
(c(1
0
) (1 `)c(1
0
+1
1
)) (7)
It can be easily seen that for most reasonable parameters this will hold. In
fact, even if 1
M
= 1. (6) and (7) can still hold provided balance sheet costs are
15
large enough. However, since we focus on primarily on situations with large balance
sheet costs and non-trivial IOER, for the remainder of the paper we will make the
following assumption that IOER is suciently large compared to the size of the
shock:
1
M
1 +
`
2
(8)
This is a sucient condition which guarantees that 1
D1
< 1
D0
in both proposi-
tions 1 and 2.
34
Furthermore, from equation 5 in both propositions we obtain an interesting
corollary:
Corollary 3 For both M M and M< M. when ` 0 we have that 1
B
< 1
D0
.
i.e. t = 0 deposits have a higher return than government bonds.
The intuition for this corollary is that deposits are risky, since a depositor who
must relocate gets a low return. In contrast, government bonds oer a certain return.
For the expected return on deposits to be equal to the expected return on bonds,
the return depositors get if they are not relocated must be greater than the return
on bonds. This liquidity risk premium is given by
`
2
(1
D0
1
D1
) (9)
and is precisely the expected loss on a unit of deposit in case of relocation.
4 Central Bank Tools
We now use this framework to analyze the RRP and the TDF. We assume that the
MMF can invest in RRPs in addition to government bonds. In contrast, only banks
can invest in the TDF. The TDF serves as a substitute to reserves for banks, but
does not bear the liquidity benet of reserves.
Both RRPs and the TDF substitute for reserves on the balance sheet of the
central bank one for one. Thus, these assets essentially absorb reserves.
34
It may be of use discussing why we could have R
D1
> R
D0
: The reason is that in t = 0
the bank loses an expected amount of
R
M
2
from the potential shock, for every additional unit of
deposit taken on. Thus, they must be compensated for this through a lower deposit rate in t = 0.
This extra cost does not exist in t = 1 since in our model there is no shock after t = 1: However,
when the shock size is small, or balance sheet costs are very large, the extra cost incurred in t = 0
is likely to be irrelevant.
16
4.1 Overnight RRPs: Fixed-Quantity vs. Fixed-Rate, Full-
Allotment
The purpose of an ON RRP is to oer a short-term investment that is available
whenever needed. The ON RRP is oered in t = 1 and allows the MMF to purchase
additional assets. This, in turn, allows relocated households to purchase MMF
shares as an alternative to redepositing in the bank at their new location.
We do not consider RRPs at date 0 to focus instead on the role of ON RRP in
mitigating the liquidity shock. We perform the analysis in the case of no interbank
lending (M M).
Proposition 4 considers the case of a xed-quantity operation (ON FQ RRP).
The central bank oers a perfectly inelastic supply of RRPs (11
FQ
) at a market
determined, perfectly competitive rate (1
FQ
).
Proposition 4 For 11
FQ
`1
0
we have in equilibrium that 1
FQ
= 1
D1
, 1
1
=
`1
0
11
FQ
. Furthermore, in this equilibrium 1
D0
, 1
D1
, and 1
B
are all higher
than the corresponding rates in proposition 1.
Proof. If 1
FQ
1
D1
, demand for RRPs exceed supply. If 1
FQ
< 1
D1
, demand for
RRPs is zero and the market does not clear. Neither of these are possible equilibria,
hence 1
FQ
= 1
D1
in any equilibrium. This directly implies that households are
indierent between re-depositing in the unshocked bank and investing in the MMF
at t = 1, and therefore deposit market clearing implies the expression for 1
1
. Then,
1
1
< `1
0
implies that c(1
0
+ 1
1
) < c((1 + `)1
0
), the balance sheet cost for the
unshocked bank in proposition 1. Item 4 in proposition 1 implies that the long- and
short-term deposit rates both increase relative to the equilibrium without RRPs
due to the decrease in balance sheet costs, and item 5 implies that the bond rate
increases accordingly.
The ON FQ RRP increases the return received by relocated households because
it decreases the balance sheet costs of the bank in the region to which relocated
households move.
Proposition 5 shows that a xed-rate, full-allotment ON RRP (ON FRFA RRP)
can achieve the same allocation as an ON FQ RRP. The ON FRFA RRP oers
an interest rate 1
FR
at t = 1 for any quantity demanded. The rate 1
FR
is set
exogenously by the central bank.
Proposition 5 If the central banks sets 1
FR
= 1
FQ
from proposition 4, we have
that 11
FR
= 11
FQ
, 1
FR
= 1
FQ
= 1
D1
, and 1
1
= `1
0
11
FQ
= `1
0
11
FR
.
17
Proof. Equality of the rates is immediate from proposition 4. Suppose that
11
FR
< 11
FQ
, and denote by 1
1;FR
and 1
1;FQ
the corresponding re-deposit vol-
umes. Date t = 1 market clearing implies that 1
1;FR
1
1;FQ
, which in turn
implies 1
D1;FR
< 1
D1;FQ
and hence 1
D1;FR
< 1
FR
= 1
FQ
= 1
D1;FQ
, a contradic-
tion. An analogous argument for 11
FR
11
FQ
implies that 11
FR
= 11
FQ
in
any equilibrium. Equality of the t = 1 deposit volumes follows from market clearing.
Figure 5 illustrates the relationship between the two RRP policies.
See gure 5
The ON FRFA RRP policy that sets 1
FR
= 1
D0
is of note in this model. This
policy eliminates the wedge between the bond and deposit rates in t = 0.
Proposition 6 If the central bank sets 1
FR
= 1
D0
then 1
B
= 1
D0
in t = 0.
Proof. From proposition 5, 1
FR
= 1
D1
in any equilibrium. Hence, 1
FR
= 1
D0
implies 1
D1
= 1
D0
= 1
B
by equation (5) of proposition 1.
We can see from the previous two propositions that the RRP creates a transfer
from the non-shocked bank to the shocked household by absorbing some of the
liquidity shock. As a result the equilibrium t = 1 deposit rate is increased, up to
the facility rate, and the quantity decreased, by the size of the facility. The overall
result of this is a decrease in bank prots (because a lower amount of t = 1 deposits
are issued at a higher rate) and an increase in consumer wealth (because shocked
funds now yield a higher return).
Note that the presence of the ON RRP indirectly exerts upward pressure on
bond rates in t = 0. The increase in the re-deposit rate for a shocked household
increases the overall expected return of investing in deposits. Arbitrage then requires
the bond and t = 0 MMF return to increase as well.
4.1.1 Uncertainty in Shock Size
While proposition 5 shows that a ON FQ RRP can implement the same allocation
as an ON FRFA RRP, it is worth thinking about how the two tools could dier in
a richer setting. For example, the two facilities would have dierent implications if
the fraction of household that are relocated, `, is uncertain. In such a case, an ON
FQ RRP would result in uctuations in the RRP rate, while an ON FRFA RRP
would result in uctuations in the quantity of RRPs. Hence, a policymaker who
18
dislikes uctuations in the interest rate more than uctuations in the quantity of
reserves would prefer the ON FRFA RRP.
To formalize this, we assume in this section that ` can take two values, `
L
and
`
H
. with `
H
`
L
. We assume that the central bank knows the two possible
realizations of ` but does not know which one will occur when it implements the
ON RRP policy.
35
We also assume that the central bank would like to target a
specic t = 1 investment rate of 1

and can choose either an ON FRFA RRP or

an ON FQ RRP to do so. We will show that, in general, the ON FRFA RRP can
implement a t = 1 investment rate close to 1

with less interest rate volatility than

the ON FQ RRP.
To make the problem interesting, we analyze the case where:
1
M
1c(1
0
) 1

1
M
1c((1 +`
L
)1
0
) (10)
so that the target t = 1 investment rate is higher than the outcome that would
occur in either state without central bank intervention, but not so high that t = 1
deposit markets become completely inactive in all situations. We rst show that an
ON FRFA RRP policy can implement 1

in either state.
Proposition 7 If the central bank sets 1
FR
= 1

. and (10) holds, then 1

D1
=
1
FR
= 1

when either `
L
or `
H
occurs.
Proof. First assume `
L
occurs. Suppose 1
D1
1
FR
. Then there would be no
demand for the ON FRFA RRP, and 1
1
= `
L
1
0
. However, since we have that 1
D1
1
FR
1
M
1c((1 + `
L
)1
0
) by assumption, the unshocked bank will not be
willing to supply `
L
1
0
in t = 1 deposits which is inconsistent with market clearing.
Thus, we cannot have 1
D1
1
FR
. Now suppose that 1
D1
< 1
FR
. Then there will be
no demand for t = 1 deposits (1
1
= 0). However since 1
FR
= 1

< 1
M
1c(1
0
).
banks will want to supply positive deposits, which is also inconsistent with market
clearing. Therefore, we cannot have 1
D1
< 1
FR
. Thus, we have established that
1
D1
= 1
FR
= 1

when `
L
occurs. Now suppose that `
H
occurs and that 1
D1
1
FR
. Since `
H
`
L
. we will have that 1
D1
1
FR
1
M
1c((1 + `
H
)1
0
) by
(10). By the same argument as in the previous part, we have that 1
D1
1
FR
is a
contradiction. The case where 1
D1
< 1
FR
is eliminated by an identical argument
as in the previous part. Thus we have that 1
D1
= 1
FR
= 1

in either state.
The proposition establishes that an ON FRFA RRP can impose a target 1

for
any state under (10) and thus completely eliminate t = 1 interest rate volatility. It
35
For the analysis we conduct here, it is actually not necessary for us to dene probabilities of
the two states occuring.
19
is rather clear that this cannot be achieved with the ON FQ RRP. This is because
when `
L
is realized, a smaller value facility will be needed to impose 1

than when
`
H
occurs. A central bank cannot achieve 1

for all realizations of ` if it cannot

condition the policy on the state of the world. We do however show that the central
bank can implement a oor on rates at 1

. First we dene 11
FQ

to be the quantity
of the facility that is needed to impose 1

when `
H
occurs, i.e. 11
FQ

will solve:
1

= 1
M
1c((1 +`
H
)1
0
11
FQ

) (11)
Such an 11
FQ

will exist by continuity of the cost function as well as (10).

Proposition 8 Under (10), if the central banks sets 11
FQ
= 11
FQ

, then 1
D1
=
1

when `
H
occurs, and 1
D1
1

when `
L
occurs, and 1
1
is positive.
Proof. The fact that 1
D1
= 1

when `
H
occurs is true by the denition of 11
FQ

.
When `
L
occurs two situations can arise. First we may have that 11
FQ

`
L
1
0
.
In such a case we will have
1
D1
= 1
FQ
= 1
M
1c((1 +`
L
)1
0
11
FQ

)
1
M
1c((1 +`
H
)1
0
11
FQ

= 1

.
However, when 11
FQ

`
L
1
0
, the facility supply cannot be satised by shocked
withdrawals alone. In order to satisfy the full amount of 1Q

additional withdrawals
must occur by non-shocked investors. This will only happen when 1
FQ
= 1
D0
so that non-shocked depositors are indierent between holding their deposits and
withdrawing and investing in the facility. Since the non-shocked bank will want to
supply zero deposits in t = 1 at 1
D0
. we will have 1
1
= 0. and the market for t = 1
deposits will not even exist in this case. Thus, whenever positive t = 1 deposits
exist, we will have that 1
D1
= 1
FQ
1

.
This proposition shows that a ON FQ RRP can eectively provide a oor on
rates at the target rate 1

. However, t = 1 investment rates may be very volatile,

especially if the dierence between the two shock sizes are large and both occur with
high probabilities. A central bank with the intention of implementing a target rate
while minimizing interest rate volatility may thus prefer a ON FRFA RRP over an
ON FQ RRP.
4.1.2 Discussion
Other advantages of a FRFA RRP may also exist that are outside the scope of this
model. For example, xed-rate RRPs provide MMFs with certainty regarding xed-
rates, and certainty regarding (unlimited) quantities, both of which would have
20
additional benets to MMFs in the face of uncertainty on demands and supplies
in short-term money markets. In practice, MMFs have eective risk aversion, in
part caused by the requirement for stable net asset values (NAVs). The certainty
provided by ON FRFA RRPs creates a benet for a more stable transmission of
monetary policy. FQ RRPs, in contrast, will not eliminate the uncertainty regarding
equilibrium rates and quantities that MMFs can receive.
Furthermore, xed-rate RRPs tend to better facilitate an overnight RRP facility.
As shown above, such daily availability provides MMFs with greater certainty to
support their elastic demand, at or above the RRP rate, for other assets. Fixed-rate
RRPs also allow for one-day maturity RRPs, which can be rolled over. A xed-
quantity RRP is not as amenable to daily operations. FQ RRPs would therefore
tend to require longer term RRPs for operational cost reasons. Shorter-term RRPs
provide MMFs the ability to substitute with shorter-term alternative assets. Such
RRPs will support money market and bank deposit rates of all maturities, even as
short as overnight rates. Typically, money market rates increase with the tenor of
the instrument. While ON FRFA RRPs can be rolled over, and therefore provide a
better oor to overnight rates as well as to longer-term rates, a longer-tenor RRP
does not provide such support to rates of shorter tenors.
As a result of this, longer-term RRPs may possibly have little eect on shorter-
term rates. For example, a one-month RRP may increase one-month money market
rates, but this may not be well transmitted down to provide support for overnight
rates, which could be best supported with ON RRPs. One-month RRPs may rather
simply increase the steepness of the one-month yield curve. In the case of ON FRFA
RRPs, MMFs may even take up quantities of RRPs at quite low rates.
4.2 Term vs. Overnight RRPs
We model long-term RRPs as an alternative source of investment for MMFs in
t = 0. Term RRPs (denoted 11
TM
) are supplied inelastically by the central bank
at a xed-rate 1
TM
. paid o in t = 2. Term RRPs are available only to the MMF
and reduce the quantity of reserves.
We restrict our analysis to the case where the interbank market is inactive at
date 1, as in proposition 1. If 11
TM
RRPs are issued in t = 0, the total supply
of reserves decreases from M to M
0
where M
0
= M 11
TM
. So we assume that
M
0
M.
Item 5 of proposition 1 must hold for bonds to be held in equilibrium. In the
benchmark case we can write the equilibrium bond rate as:
21
1
B
= (1
M
)
2

1
2
(c(1
1 M
2
) + (1 +`)c((1 +`)(1
1 M
2
)) (12)
If term RRPs are introduced at a rate below 1
B
, then no RRPs are held by the
MMF. Therefore we have the following lemma:
Lemma 9 If 1
TM
is set less than 1
B
in proposition 1, we will have
1
TM
< 1
F0
= 1
B
= 1
B
=
`
2
1
D1
+ (1
`
2
)1
D0
< 1
D0
and 11
TM
= 0. M= M
0
.
If 1
TM
is set above 1
B
, then there will be a positive demand for term RRPs.
Proposition 10 For 1
TM
greater than 1
B
, in equilibrium we must have 1
B
<
1
B
= 1
TM
, 11
TM
0. and M
0
< M. Furthermore, both 1
D0
and 1
D1
will
increase, while their respective equilibrium quantities decrease.
Proof. See the Appendix
RRP holdings increase when the term RRP rate is set at a higher level. This
leads to higher MMF investment and lower bank deposits, reducing the balance
sheet costs. Hence, both 1
D1
and 1
D0
increase until 1
TM
=

2
1
D1
+ (1

2
)1
D0
.
The term RRP rate, when it is set suciently high, creates a oor for the bond and
deposit rates. The appendix provides a proof of proposition 7. It also characterizes
the acceptable range of central bank-set RRP rates so that equilibrium is possible
without triggering an interbank market. Figure 6 graphically illustrates the aect
of the term RRP policy on 1
D0
, and the aect on 1
D1
is similar.
See gure 6
We have modeled term RRPs as being xed-price, full-allotment. As in the
previous section, one can also consider an operation for a xed quantity, 11
TM
,
of RRP, where the rate is market determined. The equilibrium in the xed-price,
full-allotment case where 1
B
= 1
B
= 1
F0
= 1
TM
is identical to any equilibrium in
which a quantity 11
TM
of RRPs are set which yield 1
TM
in equilibrium. The main
dierence is that in the xed-rate setting we could have equilibria where 1
TM
< 1
B
.
In this situation, MMFs only hold bonds and no RRPs are held. Such an equilibrium
(where the RRP rate is strictly below the other rates) is impossible in the operation
22
style setting of the xed-quantity RRP market.
36
Arbitrage forces the RRP rate to
be equal to the bond rate, since it is also a safe asset. For similar reasons, the bank
deposit rates increase with the RRP quantity supplied. The mechanism is that an
increase in the quantity supplied of RRPs decreases the quantity of deposits which
reduces balance sheet costs.
Both the term and the ON RRP increase t = 1 and t = 2 deposit rates. However,
they do so through dierent mechanisms. The ON RRP (that is oered in t = 1
only) sets a reservation t = 1 deposit rate, which directly raises 1
D1
up to that level.
1
D0
increases because the ON RRP absorbs shocked withdrawals and lowers t = 1
expected balance sheet costs for the bank. The term RRP (oered only in t = 0)
directly lowers balance sheet costs in t = 0 and partially raises t = 0 deposit rates.
The decrease in deposits at t = 0 reduces the size of the shock in t = 1. which then
indirectly increases t = 1 deposit rates by reducing the balance sheet cost burden.
This then feeds back into t = 0 deposit rates, as banks now expect lower t = 1
deposits. Thus, an additional increase in 1
D0
occurs.
An ON RRP (of either xed-rate or xed-quanity) oered at t = 0 would also
reduce 1
0
and thus reduce balance sheet costs and raise t = 0 deposit rates. Term
RRPs may be preferable, however, if oering RRPs of open-ended size on a daily
basis is operationally dicult or expensive compared to a term RRP that attracts
large, long-term deposits.
4.3 TDFs vs. RRPs
As shown in the previous sections, RRPs provide a tool for the central bank to both
manage reserves and provide a oor for rates. Another such tool is the term deposit
facility (1). The TDF is comparable to the term RRP with the exception that it is
only oered to banks. As we will show in this section, the two tools vary in their
eect on equilibrium rates depending on various parameters in the model. For this
section, we assume that the central bank chooses a suciently large equilibrium
quantity of the TDF or RRP such that reserves are reduced to the point where
M < M at the end of t = 0. That is, the central bank is using its policy tools
in large enough size so as to rekindle the interbank market and promote interbank
lending in t = 1.
We model the TDF as a xed-quantity operation (1) oered by the central bank
in t = 0 maturing in t = 2 with a competitive return 1
T
. Important to note is that
36
One could argue that this case corresponds to a zero quantity auctioned, where the equilibrium
rate i indeterminate within a range.
23
TDF holdings cannot be used to ward o liquidity shocks. In this sense, they are
a perfect substitute for real sector bank lending and in equilibrium we must have
that 1
T
= 1
L
. The key dierence between the RRP and the TDF is that TDFs
force substitution of liquid reserves to illiquid TDF holdings completely within the
banking sector. RRPs, on the other hand, divert reserve holdings to the assets
held by the universal MMF which is not prone to shocks. Thus, utilizing RRPs as
opposed to TDFs reduces liquid assets while bearing less of an increase on interbank
borrowing costs and liquidity premia. For simplicity and for ease of comparison, we
will assume that the RRP is a term RRP in this case.
As an illustration, rst suppose that the central bank increases the TDF supply
by 2 units. Each bank would decrease its liquid reserve holdings by 1 unit. In doing
so they lose 1
M
units of reserves which could have been used to ward o a potential
liquidity shock. This means that interbank loans will increase by 1
M
. which implies
that the equilibrium interbank rate will increase by ,(1 + 1
M
) ,
0
(1). On the
other hand, if the central bank were to supply 2 more units of RRPs instead, each
bank would still decrease its liquid reserve holdings by 1 unit (assuming the MMF
borrows equally from both sectors) sacricing 1
M
in liquid assets in t = 1. However,
each household is also converting one deposit to an MMF share. Thus, interbank
loans will only increase by 1
M
` < 1
M
and the interbank rate will increase by
,(1 +1
M
`) ,
0
(1) < ,(1 +1
M
) ,
0
(1). This implies that the liquidity premium
will be smaller in the case of RRPs than of the TDF. This is shown in gure 7.
See gure 7
This dierence also factors into the eect on deposit rates. From part 4 of
proposition 2, we can see that the t = 0 deposit rate is increasing in interbank
lending and decreasing in equilibrium t = 0 and t = 1 deposits due to balance sheet
costs. This leads to the following proposition for interbank and balance sheet costs
that are largely convex. The proposition is more explicitly formalized and proved
in the appendix:
Proposition 11 When marginal real balance sheet costs are large relative to in-
terbank lending costs and both marginal cost functions are convex, we will have
@R
D0
@T

@R
D0
@RP
TM
. This occurs when deposits and liquid reserves are large and inter-
bank lending is small. Reversing these relationships will yield the opposite inequality.
Proof. See the Appendix
The result is driven primarily by the convexity of the two costs. Nevertheless,
there are relatively interesting implications of the above proposition. The sizes of
24
,(.) and c(.) will increase when interbank loans and deposits, respectively, are larger.
Therefore, when both of the cost functions are very convex, a policymaker who has
a primary goal of increasing deposit rates may want to mediate between usage of
both tools. Specically, he may want to rst use term RRPs to reduce deposit size,
then divert to implementation of the TDF after decreases in marginal balance sheet
costs diminish. On the other hand, a policymaker who seeks to absorb reserves with
a smaller eect on increasing deposit rates may want to focus on the one facility,
namely the TDF if marginal balance sheet costs are very fast increasing, and the
RRP if interbank lending frictions are instead more prominent.
5 Conclusion
In response to the 2007-09 nancial crisis and subsequent economic conditions, the
Federal Reserve engaged in large-scale lending to nancial institutions to provide
liquidity and large-scale asset purchases to stimulate the economy by lowering inter-
est rates. As a result of these policies, the amount of reserves in the banking system
increased dramatically to over $2 trillion as of September 2013. In October 2008,
the Federal Reserve began oering IOER to DIs to better control short-term rates
in this environment of large excess reserves. Furthermore, in June 2011, the FOMC
announced plans to implement certain tools, namely term and ON RRPs oered to
a wide range of counterparties and the TDF oered to DIs, which would help IOER
in supporting interest rates. In August 2013, the FOMC announced further study
of xed-rate, full-allotment ON RRPs as a potential tool. The purpose of this paper
is to analyze the eectiveness of IOER, the RRP, and the TDF in implementing the
FOMCs goals of retaining control over short-term interest rates while managing
reserves.
We introduce a simple general equilibrium model to analyze these tools. While
parsimonious and tractable in its exposition, our model highlights numerous impor-
tant results regarding IOER, RRPs, and the TDF. As in MMS, IOER inuences
deposit rates as it represents the riskless return on holding a deposit. It also sets
a short-term reservation rate in the interbank market at which DIs should not lend
below. We nd that RRPs work mainly through the household investment side but
also serve to reduce reserves. RRPs can raise deposit rates in two ways: Decreasing
balance sheet costs through reallocating depositor funds into MMFs holding gov-
ernment bonds, which are not subject to balance sheet costs and liquidity shocks,
and setting a reservation rate for short-term deposits that occur as a result of liq-
25
uidity shocks. While the model shows that xed-rate and xed-quantity ON RRPs
would be identical in the absence of informational frictions, xed-rate RRPs may be
advantageous for a central banker trying to set a oor on short-term rates without
perfect knowledge on the intensity of shocks or exogenous costs agents face. Term
RRPs in this model oer little benet over ON RRPs if the ON RRP can be oered
on a consistent basis at low cost in implementation. However, they can further
reduce balance sheet costs and raise deposit rates. TDFs, in contrast to RRPs, do
not reduce balance sheet costs. Therefore, when the banking system is forced to
trade on costly interbank markets, RRPs increase the size of this interbank market
by less than the TDF. This motivates usage of the RRP and the TDF concurrently.
Namely when real marginal balance sheet costs and interbank lending costs are very
convex, policy makers with the intention of raising deposit rates to high levels may
want to use both facilities together.
Our model provides a broad framework in which many additional questions about
central bank policy for managing large levels of reserves can be analyzed. As in
MMS, we can ask how balance costs aect ination and consumption. Additionally,
we can consider heterogeneity across and within sectors. Other extensions to the
model can include incorporating additional nancial institutions, such as securities
dealers and GSEs. The central bank can manage its liabilities facilities with regard
to these institutions to further absorb reserves, inuence the broad range of money
market rates, and impact the level of lending, output, ination, and consumption.
With this in mind, we also stress the versatility of this particular model as a useful
benchmark that can be used to analyze a variety of further institutional details for
nancial intermediaries and money markets.
26
References
[1] Ashcraft, Adam, James McAndrews and David Skeie (2011). Precautionary
Reserves and the Interbank Market, Journal of Money, Credit and Banking,
Vol. 43(7, Suppl.), pp. 311 - 348.
[2] Bech, Morten, Elizabeth Klee (2011). The Mechanics of a Graceful Exit: In-
terest on Reserves and Segmentation in the Federal Funds Market, Journal of
Monetary Economics, Elsevier, Vol. 58(5), pp. 415-431.
[3] Beckner, Steven K. (2009). Federal Reserve State of Play, imarketnews.com
November 12, 2009. http://imarketnews.com/node/4617.
[4] Bennett, Paul and Stavros Peristiani (2002). Are U.S. Reserve Requirements
Still Binding? Federal Reserve Bank of New York Economic Policy Review
8(1).
[5] Ennis, Huberto M. and Todd Keister (2008). Understanding Monetary Policy
Implementation, Federal Reserve Bank of Richmond Economic Quarterly 235-
263.
[6] Freixas, Xavier, Antoine Martin and David Skeie (2011). Bank Liquidity, In-
terbank Markets and Monetary Policy, Review of Financial Stuides, 11(1), pp.
104-132.
[7] Gagnon, Joseph, Matthew Raskin, Julie Remanche, and Brian Sack (2010).
Large-Scale Asset Purchases by the Federal Reserve: Did They Work?, Fed-
eral Reserve Bank of New York Sta Reports no. 441.
[8] Hilton, Spence (2005). Trends in Federal Funds Rate Volatility, Federal Re-
serve Bank of New York Current Issues in Economics and Finance 11(7).
[9] Kashyap, Anil and Jeremy C. Stein (2012). "The Optimal Conduct of Mone-
tary Policy with Interest on Reserves," American Economic Journal: Macro-
economics, 4(1), pp. 266-282.
[10] Keister, Todd, Antoine Martin and James McAndrews (2008). Divorcing
Money from Monetary Policy, Federal Reserve Bank of New York Economic
Policy Review 14(2).
[11] Keister, Todd and James McAndrews (2009). Why Are Banks Holding So
Many Excess Reserves? Current Issues in Economics and Finance, Federal
Reserve Bank of New York.
[12] Martin, Antoine, James McAndrews and David Skeie (2013). Bank Lending
in Times of Large Reserves, Federal Reserve Bank of New York Sta Reports
no. 497.
[13] Meltzer, Allan H. (2010). The Feds Anti-Ination Exit
Strategy Will Fail. Wall Street Journal. January 27, 2010.
http://online.wsj.com/article/SB10001424052748704375604575023632319560448.html.
[14] Poole, William (1970). "Optimal Choice of Monetary Policy Instruments in a
Simple Stochastic Macro Model" Quarterly Journal of Economics 84(2) pp.
197-216.
27
[15] Wrightson ICAP (2008). Money Market Observer, October 6.
[16] Wrightson ICAP (2009). Money Market Observer, January 12.
28
Symbol Description
1 Supply of government bonds
1
H
Bonds held by MMF on behalf of households
1
CB
Bonds held by central bank
c(.) Marginal real balance sheet costs as a function of deposits
1
0
Deposits oered by banks to households in t = 0
1
1
Deposits oered by non-shocked bank to households in t = 1
1 t = 0 household endowment of individual sector
,(.) Marginal real interbank lending cost as a function of interbank loans
1
0
MMF shares held by household in t = 0
1
1
MMF shares held by household in t = 1
1 Interbank loans in t = 1
1 Bank loans to rms in t = 0
M Total reserves supplied by central bank
` Reserves held by an individual bank
M Threshold level of reserves supply at which any less would trigger the interbank market
1 t = 2 price of real consumption
:(.) Marginal rate of rm production as a function of bank loans
1
B
Return on government bonds paid at t = 2
1
D0
Return on t = 0 deposits paid at t = 2
1
D1
Return on t = 1 deposits paid at t = 2
1
F0
Return on t = 0 MMF shares paid at t = 2
1
F1
Return on t = 1 MMF shares paid at t = 2
1
I
Return on interbank loans paid at t = 2
1
L
Return on t = 0 loans paid at t = 2
1
M
Exogeneously set return on reserve held for 1 period
` Size of the relocation shock as a fraction of deposits
29
Appendix Proofs:
Proof of Proposition 1. We show that if M M, then there exists an
equilibrium (Q,R) given by proposition 1, equations (1)-(8). This equilibrium is
unique up to the allocation of bond holdings between MMFs and households and
the return on interbank loans 1
I
1
M
. For notational convenience, we suppose
that MMFs are the only private holders of bonds whenever there is no ambiguity.
To begin, we show that the proposed price system and allocation is an equilib-
rium. Equation (5) follows from the households rst order conditions and implies
that they are indierent between holding shares in MMFs and depositing funds in
expectation at t=0. The equality of the MMF and bond returns follows imme-
diately from the assumption of competitive pricing, and implies that the MMF
is indierent about the number of shares it issues. Hence 1
0
=
BM
2
is opti-
mal, and is necessary for bond market clearing. Equation (2) follows from the
banks rst order condition and implies that banks are indierent between holding
loans or reserves on the asset side of their balance sheets. Equation (4) follows
from indierence conditions about balance sheet size for the banks. In periods
t=0 and t=1, the return from a marginal reserve must equal the cost of a mar-
ginal deposit in expectation. (In t=1, this indierence only need apply to the
unshocked bank, by assumption.) The indierence condition in t=0 is given by
0 =
1
2
((1
M
)
2
1
D0
1c(1
0
+1
1
)) +
1
2
((1
M
)
2
(1 `)1
D0
`1
M
1c(1
0
)) and
in t=1 it is 1
D1
+ 1c(1
0
+ 1
1
) = 1
M
, where 1
0
and 1
1
are determined endoge-
nously. Rearranging yields the rst two equations in (4). Given the nal equation
in (4), 1
1
= 0 is optimal for households. Binding budget constraints then directly
implies 1
1
= `1
0
. Then 1
0
= 1
BM
2
is necessary to satisfy the households
budget constraints and, given the banks indierence between loans and reserves at
t=0, directly implies (1). By the assumptions that :(.) 1 and that 1
M
0 is
set exogenously, there exists a positive 1 satisfying (8). Assumptions on :(.) imply
that rms always demand 1 0, and given (8), (1) satises their rst order con-
ditions. Finally, given (7), unshocked banks in t=1 are indierent between lending
and keeping a marginal reserve, so (6) is clearly optimal. Thus, this allocation is an
equilibrium under the given price system.
To show uniqueness, we argue that, aside from the possibility of 1
I
< 1
M
,
these prices must hold in any equilibrium and that 1 = 0 in any equilibrium. The
no-arbitrage arguments for (2), (4), and (5) show that these must hold in any equi-
librium. To be explicit, (2) is required so that there is not innite (zero) supply of
loans and zero (innite) demand for reserves; (4) is required for banks to demand
positive, nite balance sheets; (5) is required for nite, positive household demand
30
for both deposits and bonds. If 1 does not satisfy (8), either the rms rst order
conditions are not satised or one of the t=0 markets does not clear, so 1 must
satisfy (8) in any equilibrium. To nish, we consider the household and bank port-
folio decision at t=1. Suppose 1
F1
1
D1
. Then it is optimal for liquidity-shocked
households to invest all of their withdrawn funds in the MMF such that 1
1
= 0
and 1
1
= `1
0
. In this case, the shocked bank holds
M
2
`1
0
reserves and the
unshocked bank holds
M
2
reserves (its volume of deposits is unchanged after the
shock), so the market for reserves does not clear in t=1, a contradiction. So the
only admissible equilibrium t=1 rates satisfy 1
F1
1
D1
. Similarly, 1
I
1
M
is
necessary for banks to have nonzero demand for reserves at t=1. If inequality is
strict, no interbank loans will be issued because holding reserves strictly dominates
lending them. Suppose the weak inequality binds and that (6) does not hold, that
is, 1 0 in equilibrium. In this large-reserves regime, this would imply violations
of both banks t=1 budget constraints. Hence the shocked bank must lend back
to the unshocked bank, and 1 0 cannot be an equilibrium. Hence, the proposed
equilibrium is the unique symmetric equilibrium up to the return on interbank loans
1
I
1
M
.
Proof of Proposition 2. We showthat if M< Mand :(1
B
2
)
1
2
1
M
,(`(1
BM
2
) 1
M M
2
), then there exists a unique equilibrium (Q,R) given by proposition
2, equations (1)-(8).
To begin, we show that equations (1)-(8) constitute an equilibrium. Equation
(2) is a direct generalization of its analogue in proposition 1, and implies banks
indierence (in expectation) about holding loans or reserves as assets at t=0. The
indierence condition is 0 =
1
2

1
L
1
M
(1
I
1,(1))

+
1
2
(1
L
1
M
1
I
), where the
rst term is the unshocked banks return on a marginal loan net the opportunity
cost of holding a marginal reserve to be lent out, and the second term is the shocked
banks return on a marginal loan net the realized cost of not holding a marginal
reserve. Similarly, Equation (4 ) is derived from banks indierence about holding
a marginal reserve and deposit at t=0, which is expressed as 0 =
1
2
(1
M
1
I
1
D0

1c(1
0
+1
1
)) +
1
2
((1
M
)
2
(1 `)1
D0
`1
I
1c(1
0
)). Now, (7) implies that the
unshocked bank is indierent at t=1 between holding or lending a marginal reserve,
when it has already lent 1. Hence, (6) is optimal given (7) and also ensures market
clearing for deposits at t=1. Consider 1 is given by 1 =
R
L
r(E
B
2
)
. Expanding 1
L
and rearranging shows that this is equivalent to 1 =
(R
M
)
2
r(L)
1
2
R
M
f(I)
when 1. 1
L
and
1 assume their equilibrium values given by (1), (2) and (7). The arguments in the
proof of proposition 1 show that equations (1), (3), and (5) follow from (2) and (4),
31
and are also consistent with rm optimization given (8). Now, (7) implies that the
unshocked bank is indierent at t=1 between holding or lending a marginal reserve,
when it has already lent 1. Hence, (6) is optimal given (7) and also ensures market
clearing for deposits at t=1. Hence, (1)-(8) dene a competitive equilibrium.
To show uniqueness, it suces to show that the price system must hold in any
equilibrium. The above arguments then imply that the allocation is the unique
symmetric equilibrium. The price systemmust hold in any equilibriumby arguments
identical to those in the proof of proposition 1 in addition to the observations that,
rst, (7) must hold in order for interbank loan supply to be positive and nite,
and second, (6) must hold exactly to satisfy the banks t=1 budget constraints
and clear the market for t=1 deposits. Hence, (1)-(8) dene the unique symmetric
equilibrium.
Proof of Proposition 10 We will rst introduce some notation and then restate
and prove the proposition. First, note that the equilibrium t = 0 and t = 1 deposit
rates can be written as a function of the amount of term RRPs that is invested in by
the household of each sector through the MMF. Denote the functions 1
D0
(11
TM
)
and 1
D1
(11
TM
) respectively.
37
Now, recall that term RRPs absorb reserves. Thus,
we dene
\
11
TM
=
(MM)
2
as a threshold level of total term RRPs contributed by
the endowment of each sector so that if any additional units were supplied, the
interbank market would be triggered.
Furthermore, we dene
d
1
D0
and
d
1
D1
as the equilibrium rates that would result
if 2
\
11
TM
and M were supplied by the central bank:
d
1
D0
=
2
2 `
[(1
M
)
2

`1
M
2
1(
1
2
c(1
1
2

M
2
) +
1
2
c((1 +`)(1
1
2

M
2
))]
d
1
D1
= 1
M
1c((1 +`)(1
1
2

M
2
))
Finally we dene
[
1
TM
as:
[
1
TM
=
`
2
d
1
D0
+ (1
`
2
)
d
1
D1
Now we restate the proposition:
Restatement of Proposition 10: For

2
1
D0
+ (1

2
)1
D1
< 1
TM

[
1
TM
.
where 1
D0
and 1
D1
are the equilibrium rates given in proposition 1, there exists a
37
In this proof we denote RP
TM
as the amount of each households endowment that goes to
funding the MMFs investment in term RRPs. It is equal to half of the total amount of term RRPs
supplied by the MMF.
32
competitive equilibrium where 1
B
= 1
TM
and 11
TM
0. where 11
TM
is the
term RRP quantity consumed by each sector. Furthermore, 1
D0
and 1
D1
rise to
1
D00
1
D0
and 1
D10
1
D1
. and 1
0
and 1
1
reduce to 1
0
0
and 1
1
0
. All other
rates and quantities remain unchanged. This is the only equilibrium in which the
interbank market is not triggered.
Little actually remains to be proved. Note that both the equilibrium rates
1
D0
(11
TM
) and 1
D1
(11
TM
) are continuous and strictly increasing in 1` by
proposition 1, therefore the expression

2
1
D0
(11
TM
)+(1

2
)1
D1
(11
TM
) is contin-
uous and strictly increasing in 11
TM
. Also

2
1
D0
(0) +(1

2
)1
D1
(0) =

2
1
D0
+(1

2
)1
D1
< 1
TM
by assumption, and

2
1
D0
(
\
11
TM
) + (1

2
)1
D1
(
\
11
TM
) =
[
1
TM

1
TM
. Intermediate value theoremmandates that there exists a 0 < 11
TM
<
\
11
TM
such that

2
1
D0
(11
TM
) + (1

2
)1
D1
(11
TM
) = 1
TM
. 11
TM
is our equilib-
rium Term RRP consumption for each sectors. Clearly 1
B
= 1
TM
for equilibrium
in the bond market. 1
D0
and 1
D1
have increased to 1
D00
= 1
D0
(11
TM
) and
1
D10
= 1
D1
(11
TM
) respectively. 1
0
and 1
1
reduce to 1
0
0
= 1
B
2
+(
M
2
11
TM
)
and 1
1
0
= `1
0
0
respectively.Uniqueness follows immediately from the fact that the
expression

2
1
D0
(11
TM
) + (1

2
)1
D1
(11
TM
) is strictly increasing
Proof of Proposition 11
38
: We rst rewrite the 1
D0
of proposition 2 in terms
of the quantity of term RRPs (11
TM
) and TDF (1) invested in by the sectors.
Note that in this case 1
0
= 1
B
2
+
M
2
11
TM
, 1
1
= `1
0
. Also we denote `
0
as
the eective level of reserve holdings each sector holds after central bank issuance
of term RRPs and TDFs. That is `
0
=
M
2
11
TM
1.
1
D0
(11
TM
. 1) =
2
2 `
[(1
M
)
2

`1
M
2
+1,(`1
0
(1
M
)
1=2
`
0
)(
1
M
2

`
2
) 1(
1
2
c(1
0
)
+
1
2
c((1 +`)(1
0
)))]
We dene the following quantities:
c =
1
2 `
, = :(1)
1
2
1
M
,(`1
0
1
M
`
0
)
c = ,(`1
0
1
M
`
0
)(1
M
`) (c(1
0
) +c((1 +`)(1
0
)))
38
As in the proof of proposition 10, we denote RP
TM
and T as the amount contributed by each
sector rather than the total supply.
33
Now taking the derivatives of 1
D0
with respect to both 11
TM
and 1. algebra
will yield:
J1
D0
J11
TM
= c1(
1
M
(1
M
`),
0
(1)c
2,
+,
0
(1)(1
M
`)
2
+c
0
(1
0
) + (1 +`)c
0
((1 +`)1
0
)
J1
D0
J1
= c1(
(1
M
)
2
,
0
(1)c
2,
+,
0
(1)1
M
(1
M
`)
We can now write the expression:
J1
D0
J1

J1
D0
J11
TM
= c1(
1
M
,
0
(1)`c
2,
+,
0
(1)(1
M
`)`c
0
(1
0
)(1+`)c
0
((1+`)1
0
)
From the above equation, we can see that
@R
D0
@T

@R
D0
@RP
TM
0 when:
,
0
(1)(
1
M
`c
2,
+ (1
M
`)`) c
0
(1
0
) + (1 +`)c
0
((1 +`)1
0
)
Since it was assumed that both ,(.) and c(.) are convex, their derivatives are
both increasing in interbank and deposits respectively. Also, , is decreasing and c
increasing in the marginal interbank lending cost, so the left side of the inequality
will be increasing in the marginal interbank cost. c is decreasing in the marginal
balance sheet cost, so the left side of the inequality is decreasing in marginal balance
sheet costs. Thus, we have that the left side of the inequality will be larger when 1)
Interbank lending is high and 2) deposits are low.
34
Figure 1
Appendix: Figures
Figure 2
(A)
(B)
(C)
(A) Bond market: Household bond holdings increase leftward. As households hold more bonds, deposits decrease in equilibrium.
Deposit rates are less depressed by balance sheet costs, and the equilibrium bond rate rises.
(B) Deposit market: Bond market clearing and the binding resource constraint forces household supply of deposits to be perfectly
inelastic. Banks never offer a deposit rate higher than IOR, else they would be making negative profit on marginal deposits.
Banks demand for deposits are decreasing in the deposit rate.
(C) Loan market: Banks loan supply correspondence is zero below IOR, infinite above IOR, and perfectly elastic at IOR. Firms
demand for loans is decreasing in the loan rate.
Figure 3

Interbank
Loans
I
n
t
e
r
b
a
n
k

R
e
t
u
r
n

l
0
(

)
1/2

)
1/2

0
Interbank market: When the interbank market is inactive, arbitrage forces the interbank rate to equal IOR. For positive volumes of interbank
trade, lending frictions force a wedge between IOR and the interbank rate. The shocked bank demands exactly enough interbank funds to fully
cover the liquidity shock. The unshocked banks supply of loans is increasing in the interbank rate (which makes lending more profitable).
Figure 4

Loans
L
o
a
n

R
e
t
u
r
n

Assume that

> . With reserves , the equilibrium at Point A is given by Proposition 1. Consider a decrease in reserves to . Point
B shows the partial equilibrium effect: the interbank market becomes active, forcing a wedge between the loan rate and IOR. As the supply
curve shifts upward with the loan rate, the new (partial) equilibria are traced by the (partial equilibrium) demand curve. Interbank lending
frictions also force inflation (an increase in ) as in Equation 7 of Proposition 2. Inflation shifts out the demand for loans in general
equilibrium, which results in the final (general) equilibrium at Point C. Note that the volume of loans is unchanged, as it must be due to
resource constraints.

Figure 5
D
e
p
o
s
i
t

R
e
t
u
r
n

(

)
1/2

0

l
0

1

(
0
+l
0
)
(
0
+
1
)

1

l
0

(
0
+l
0
)
(
0
+
1
)

)
1/2

1

0
Deposits
(A)
(B)
(A) Deposit market with fixed-quantity RRPs: Deposits supplied by households mechanically decrease by the amount of RRPs supplied, which forces the deposit rate
up to equal the market-determined RRP rate. The presence of RRPs decreases balance sheet costs, thereby decreasing the spread between the deposit rate and
IOR.
(B) Deposit market with fixed-rate RRPs: The deposit rate increases to match the exogenous RRP rate and market clearing forces household deposit supply to
decrease. As in (A), smaller balance sheet costs decrease the spread between the deposit rate and IOR.
Note: Figures shown with equal volumes of RRPs to demonstrate Proposition 5.
Figure 6
Deposit market with daily RRPs: The presence of RRPs forces down the household supply of deposits one-for-one with reserves. Fewer deposits imply
smaller balance sheet costs, which decreases the spread between the deposit rate and IOR.
D
e
p
o
s
i
t

R
e
t
u
r
n

Figure 7
L
o
a
n

R
e
t
u
r
n

Loans
0

1
2
()
( +

1
2
l) ()

0

(A)
(B)
(A) Loan market with TDF: The TDF soaks up reserves from the banks, activating the interbank market and driving a wedge
between the loan rate and IOR. Households supply the same quantity of deposits.
(B) Loan market with term RRPs: The RRPs force down household deposit supply one-for-one with reserves. The drop in
reserves drives a wedge between the loan rate and IOR, but the drop in deposits dampens the required volume of
interbank lending relative to (A). Hence, the spread between the loan rate and IOR is smaller for term RRPs.